February 2017 Summaries

Toastifai is an innovative AI-powered toaster developed by Jessie Pullaro and Frank Callas, students at Florida Polytechnic University, which employs Clarifai’s visual recognition API to achieve the perfect level of toastiness. By training a visual recognition model with images of toast at various stages, the team created a toaster capable of identifying and stopping the toasting process at the optimal moment, even sending a text alert when the toast is ready. The project, which integrates Python, openCV, Pillow, Twillio, a Logitech camera, Raspberry Pi, Arduino, and temperature sensors, was inspired by a desire to eliminate the common issue of burnt or forgotten toast. The developers praised the ease of use and low latency of the Clarifai API, highlighting its drag-and-drop features and simplicity, which greatly facilitated their work.

Distill is a Google Chrome extension designed to improve internet browsing by allowing users to block unwanted images and text, providing a safer and more pleasant online experience. Users can customize their browsing by selecting categories of content they wish to avoid, such as spiders or blood, and the extension automatically obscures these images. Utilizing Clarifai's Custom Training technology, Distill also enables users to train models to recognize and block specific concepts, enhancing personalization. The extension is beneficial not only for general users but also for individuals recovering from trauma or diseases, such as PTSD, by reducing potential triggers. Developed primarily with JavaScript and using Clarifai's image-tagging API, Distill was created to address the need for a cleaner web experience, inspired by the team's encounters with undesirable content online.

The text provides a detailed tutorial on using Clarifai's new Face Detection model in conjunction with JavaScript to create a humorous photo filter application that overlays emojis on detected faces in images. The model, which is in its alpha release, identifies facial positions in images by returning bounding box coordinates, making it useful for various applications such as security camera footage and digital photography. The tutorial guides users through setting up a Clarifai account, initializing a new application with the JavaScript client, and creating a project that leverages file input to select images and apply the face detection model. Once faces are identified, the tutorial explains how to map the bounding box coordinates to actual image dimensions and overlay emojis, like the poop emoji, on the detected faces using HTML canvas. The process involves converting images to base-64 strings, predicting face positions with the Clarifai model, and drawing the emojis on a canvas element, encouraging users to explore creative variations by using different emojis.

Clarifai has opened a new office in San Francisco as part of its post-Series B expansion, aiming to build a prominent machine learning research team under the leadership of Andrea Frome, a former member of Google Brain and the Hillary for America campaign. The San Francisco office will focus on innovative research to productize solutions quickly for global developers, while the New York headquarters continues to expand with recent hires from Twitter to enhance visual recognition models. On the business front, Matt Molinari, the new VP of Business Development from Indeed, will spearhead partnership efforts with other platforms. Clarifai is actively hiring across various roles, inviting individuals interested in joining the machine learning revolution to explore opportunities on their careers page.